B. Mohler, B. Raffin, H. Saito, and O. Staadt (Editors)
Evaluation of Surround-View and Self-Rotation in the O CTA V IS VR-System
Eugen Dyck1,2, Thies Pfeiffer2, Mario Botsch1,2
1Computer Graphics Group, Bielefeld University
2Center of Excellence Cognitive Interaction Technology, Bielefeld University
!"#$%&'$())*
+",'!-$.
("!/(,&)*$"0)(
$"11)$!"(&(-*2 '#(3)-11/*$)&$/4)(/
Figure 1:In theOCTAVISsystem (left) eight screens are arranged in an octagon to provide a360 surround-visualization of the virtual environment. Two door segments can be opened (center). Navigation in the VR is performed through an office chair, whose orientation determines the movement direction, and a “throttle joystick” in the armrest (right).
Abstract
In this paper we evaluate spatial presence and orientation in theOCTAVISsystem, a novel virtual reality platform aimed at training and rehabilitation of visual-spatial cognitive abilities. It consists of eight touch-screen displays surrounding the user, thereby providing a 360 horizontal panorama view. A rotating office chair and a joystick in the armrest serve as input devices to easily navigate through the virtual environment.
We conducted a two-step experiment to investigate spatial orientation capabilities with our device. First, we ex- amined whether the extension of the horizontal field of view from 135 (three displays) to 360 (eight displays) has an effect on spatial presence and on the accuracy in a pointing task. Second, driving the full eight screens, we explored the effect of embodied self-rotation using the same measures. In particular we compare navigation by rotating the world while the user is sitting stable to a stable world and a self-rotating user.
Categories and Subject Descriptors(according to ACM CCS): I.3.6 [COMPUTER GRAPHICS]: Methodology and Techniques—Interaction techniques
1. Introduction
Visual-spatial abilities are a key attribute for managing ev- eryday life. No matter whether we travel by plane to a con- ference, by train to visit a relative, or by car to see the ocean, we need to locate ourselves in space to decide which direc- tion to go to arrive at a fixed destination. Also to find our way around in a building we rely on our spatial understanding.
We do so even when other people describe specific locations to us, when we interpret pictographic signs, or when we read any kind of spatial map.
Since spatial thinking is one of the seven primary mental abilities in the theory of intelligence, and also since sports, design, mechanical construction, and even perception of mu- sic strongly depend on a functioning spatial cognition, we
c The Eurographics Association 2013.
are very limited when the corresponding brain areas get in- jured (e.g., through stroke, neuro-degenerative disease, or an accident). Luckily, visual-spatial abilities can be improved or recovered through specific training.
Recent developments in virtual reality (VR) research have shown that simulated environments in appropriate devices can be used for therapy, such that training effects can suc- cessfully be transferred to real life. In a highly interdisci- plinary research effort we developed an immersive VR de- vice for clinical studies focused on visual-spatial training and rehabilitation. Our VR system is called OCTAVIS and is illustrated in Figure1.
To mimic visual perception of reality as closely as possi- ble the OCTAVISpresents the virtual environment in a 360 panorama view on eight touch-screen displays surround- ing the user. HMDs disqualified for our project because of their low acceptance by patients. A well suited interaction metaphor for VR navigation is real walking or walking in place. However, many stroke patients are not able to walk or even stand. Hence, we employ a rotating chair as input device for navigation, where the walking direction is deter- mined by the chair orientation and forward/backward move- ment is controlled by a joystick in the armrest. A detailed description of the OCTAVISsystem is given in [DZK⇤12].
While this system has already proven its effectiveness in clinical studies [ZDK⇤13], in this paper we evaluate two of its particular properties: We analyze which effect on spa- tial presence and spatial orientation skill (i) the eight-screen surround-view and (ii) the user’s physical self-rotation have, using the MEC SPQ and a custom pointing task.
2. Related work
It has been shown that spatial knowledge does not transfer easily from virtual reality to the real world [Pso95,GM98].
We want to test the effect of the OCTAVISon the ability to orient oneself and to navigate through the virtual environ- ment, the so-calledway-finding. In the initial definition of way-finding by Kevin Lynch [Lyn60] the four components of way-finding are: orientation, route decisions, mental map- ping and closure. See Raubal and Egenhofer [RE98] for an overview. Mental maps, also called cognitive maps, are the concepts agents create in their minds to enable them to plan their activities. They are an abstraction of the environment based on the sensory input of the agents. Mental maps are (sometimes) helpful in way-finding, but way-finding does not necessarily have to rely on mental maps.
In their work on child development, Piaget and In- helder [PI67] differentiate between perceptual space and conceptual space. The perceptual space is created and in- herently tied to sensori-motor activities that lead to corre- sponding perceptions. Its development precedes that of the conceptual space in early childhood. The orientation aspect of way-finding can be linked to the concept of perceptual
space, whereas mental mapping is clearly linked to con- ceptual space. So there are two different cognitive mecha- nisms that are at work in way-finding. At least the first in development, the perceptual space, is strongly grounded in sensori-motor activities and thus some effort has to be taken to provide sufficient sensori-motor cues. A VR interface for navigation has thus to provide sufficient cues for both lev- els to fully support natural way-finding, e.g., by provid- ing appropriate visual cues and a suitable locomotion tech- nique [DP02]. In the OCTAVIS, we implemented techniques to address both levels: a 360 surround-view to maximize vi- sual cues and a chair allowing physical self-rotation for the locomotion part (Figure1).
2.1. Related work on locomotion
There are effects of the choice of the locomotion technique on way-finding and in particular on orientation. It has been shown that locomotion interfaces that provide proprioceptive and vestibular feedback have a positive effect on the user’s navigation capabilities [DS93,CGBL98,RL09]. According to expectation, interfaces that come close to the real walking experience provide an excellent feedback and support navi- gation performance of humans (see e.g. [RVB11]).
Interfaces that come close to natural walking are redi- rected walking [Raz05], motion compression [NHS04], seven-league-boots [IRA07], and scaled-translational- gain [WNM⇤06]. However, all of these interfaces have been designed to allow for navigation in larger-than-real spaces (i.e., the size of the virtual world is larger than the real estate) and thus feature some kind of compression or scal- ing [WNR⇤07]. Some of these techniques require a method for re-orienting the user in the virtual world whenever she approaches a barrier in the real world, which might reduce presence [Raz05,PFW09]. On the other side, interfaces that only allow for a partial approximation of natural walking, such as walking-in-place or treadmills [Hol02,DCC97], are difficult to handle [SGF⇤10] and might thus yield only suboptimal performance.
Currently, there is a disagreement about whether real walking simulation is the key factor [RL09] or if it is suffi- cient to make the users do physical rotations to alter their ori- entation [RSPA⇤06]. Motivated by recent work of Riecke et al. [RBM⇤10] who found excellent orientation performances when pairing physical rotation with continuous joystick- based translation, we test whether this also applies for the situation in the OCTAVIS, where the user remains seated.
There is not much work on navigation through virtual re- ality in a seated position. A recent exception is the work on redirected driving by Bruder et al. [BIPS12]. They use electric wheelchairs that drive in the real world to provide sufficient proprioceptive feedback to the driving user. How- ever, they are using HMDs to provide a full visual immer- sion into virtual reality. They find that people are a little
135°
real space virtual space
rotate
rotate travel
travel
360°
real space virtual space
rotate
rotate travel
travel
360°
real space virtual space
travel travel
rotat e
Figure 2:We designed three conditions for our study. Left: The Frontal-View condition (FV) has a 135 field of view (3 displays) and rotates/travels the virtual space through a joystick. Center: The Surround-View condition (SV) employs a 360 field of view and again navigates the VR by joystick. Right: The Rotating-User condition (RU) has a 360 field of view, but the user rotates in real space to control walking direction in the virtual space; the joystick is used to navigate forward/backward only.
less sensitive to a redirection of their orientation when driv- ing the wheelchair than when walking. This could imply that users are less sensitive to rotations of the wheelchair.
In our system, however, the users have full control of their rotation, while in the study of Bruder and colleagues a joy- stick was used to guide the motor-driven wheelchair. A promising insight of their study is that they found no sig- nificant difference in presence, thus driving through the vir- tual world while being seated was sufficient to establish a similar presence as natural locomotion. However, comparing locomotion using several strategies Nybakke et al. [NRI12]
found that task performance was better using a motorized wheelchair than traveling in a seated position with trans- lation controlled by a joystick and rotation controlled by swiveling. Their results also suggest that distributing transla- tion and rotation movements over two modalities could lead to more sequential locomotion patterns.
2.2. Related work on field of view
In early work on the effect of the field of view on presence, Hendrix and Barfield [HB95] already showed that increas- ing the field of view increases presence. However, their re- stricted technical setup only allowed a comparison in a range of 10 to 90 . Interestingly, they did not find a significant improvement between 50 and 90 .
A decrease of the field of view was observed to come along with a decrease in cognitive map building perfor- mance [AM90]. McCreary and Williges [MW98] reported that the field of view did not have an effect on landmark knowledge, but had a significant effect in a pointing task, where participants had to point to occluded objects. They also found a correlation between computer experience and performance in the pointing task.
In a desktop-based setup with up to 180 , Seay et al. [SKHR01] showed that an increase of the field of view to 180 increases the feel of presence, but also increases nau-
sea, especially for passengers not actively navigating in the virtual world. Lin et al. [LDP⇤02] tested displays with fields of view of 60 , 100 , 140 , and 180 . They report an in- crease of presence with an increase of field of view, albeit the increase seemed to approximate a plateau between 140 and 180 . In their memory test, the results also correlated positively with the perceived presence.
There is a general effect of underestimating traveled dis- tance in virtual reality [WK98], which might affect way- finding. Knapp and Loomis [KL04], however, could not find a significant effect of a limited field of view on distance es- timation. Steinicke et al. [SBJ⇤10] showed for HMD-based projections that participants underestimated rotation in vir- tual reality. However, in the OCTAVISthe frames of the pro- jection screens provide fixed anchors for orientation and thus the estimation of rotation should be rather congruent.
Under the assumption that an increased field of view facil- itates orientation and thus supports cognitive map building, the OCTAVISwas designed to support the maximum field of view independent of the viewing direction of the user. This does not hold for most desktop- or projection-based studies mentioned above and the effects remain to be evaluated.
3. Method
To investigate spatial presence and orientation in the OCTAVIS we designed a two-step experiment. In the first step we modified the real field of view and in the second step we changed the rotation aspect of our navigation metaphor.
We did so comparing three experiment conditions:
Frontal View (FV): In the first condition (Figure2, left) we used three displays to render the virtual maze. This re- sulted in a wide, but still limited, frontal view of 135 . To navigate the world the user operated a joystick with two axes: one to move forward/backward and the other to rotate the virtual scene in order to change the travel-
Figure 3:Maps of the three museum-like mazes each participant saw during consecutive runs. Participants always started at the central room (black dot). Here and in the rooms at the end of each corridor an image was displayed. These four images were to be remembered in their spatial relationship to each other and pointed to during the pointing task.
ing direction. This rotating world principle is a common metaphor used in most first person desktop games.
Surround View (SV): For the second condition (Figure2, center) we kept the rotating world metaphor, but extended the field of view to a 360 horizontal surround view employing all eight displays for rendering the maze. To prevent confusion about the front-direction we explicitly marked one monitor as front-monitor.
Rotating User (RU): The third condition (Figure2, right) used the 360 surround-view rendering, but instead of a rotating virtual world with the user sitting stable, now the user physically rotated herself to control the traveling di- rection. This way, “front” was where the user turned the chair to, which was measured by a rotary encoder in the office chair (Figure1, [DZK⇤12]). The joystick had just one axis enabled used to move forward/backward.
To enable a comparison of the three conditions in a within-subject manner, three mazes have been designed. Ac- cording to the findings in the field of space syntax and re- search about dementia-friendly architecture [MS09,DE00, BWÖ04], orientation in a building benefits from straight cir- cular corridor systems, right-angled turns, architectural sym- metry, direct sight, large rooms, and different textures and geometries per room.
Since in our study we aimed at challenging orientation, we designed our mazes to be hard with respect to the above criteria: We developed a modular system of 30 equally textured small octagonal cells with different entrance-exit- combinations. These cells can be configured on a rectangu- lar grid to create a maze, a corridor arrangement, or an ar- chitectural room-assembly. We conducted a pre-study with different cell combinations to ensure both a sufficient and comparable complexity between mazes. Based on the expe- rience from the pre-study, we decided for the three corridor- arrangements shown in Figure3. Each arrangement consists of three asymmetrically coiled corridors, some of which fea- ture unusual turns of 45 .
Semantically, each maze acted as a museum hosting four images of the same topic. The first one showed how different creatures lunch (humans, bears, sharks, eagles), the second one how they move (humans, dogs, cheetahs, bears) and the last one was about styles of government (monarchy, Obama, Putin, Merkel). The images were displayed in the central room joining the three wings and also in the last room at the end of each corridor (Figure3). An impression of an inside view is given by Figure4.
Figure 4:Screenshot of the maze as participants saw it.
In the actual experiment every participant had to perform three consecutive sub-experiments, each in a different maze.
All participants started in Maze 1 and finished with Maze 3, but we randomized the three conditions (FV, SV, RU) to be used for the mazes, such that 16 participants (8~, 8|, age=23.2, sd=4.6) performed the experiment in the order FV-SV-RU, 16 (8~, 8|, age=24.5, sd=10.0) in the order SV- RU-FV, and further 17 (8~, 9|, age=23.7, sd=6.1) in the order RU-FV-SV. In total we tested 49 healthy participants (24~, 25|, age=23.8, sd=7.1) with different education back- grounds (pupils, university students, craftspeople, office em- ployees).
After the personal data acquisition including a self-rating on VR-experience (VR-skill), a short training in a circular
corridor with two pictures was performed before each sub- experiment. Here participants learned the respective naviga- tion metaphor, the viewing setup, and about their pointing accuracy. The following sequence describes the actions per sub-experiment.
1. Be spawned in the central room and memorize its picture.
2. Walk to the end of each corridor, memorize the picture in the last room, and walk back to the central room.
3. Point from the central room to each of the pictures in the corridors in the order visited.
4. Again, walk each corridor to its end, point back to image in the central room and to the images in the other corri- dors, and travel back to the central room.
5. Point from the central room to each of the pictures in the corridors in the visited order.
In case they forgot, participants were allowed to ask in which corridor a specific picture was displayed. After com- pleting the set of three experiments the participants had to fill out a questionnaire. Finally we tested the partic- ipants’ mental rotation capabilities with the paper-and- pencil Mental-Rotation-Testby Vandenberg [VK78] in its A-version (MRTA).
To measure the spatial presence and spatial orientation skill for each condition we employed two measurement in- struments:
Pointing Task: First, a pointing task was performed within the virtual maze [CGBL98]. Pointing to an object was done by touching a touch-screen, which issued a ray from the viewer’s position in VR through the touched pixel of the respective screen/view. The horizontal angular differ- ence between this ray and the exact direction to the object in question was used as the angular error.
Questionnaire: Second, a combined questionnaire was filled out. The Measurement-Effects-Conditions Spatial Presence Questionnaire (MEC SPQ) [VWG⇤04] mea- sures not only the feeling of presence in general, but fo- cuses on spatial presence. We used the four-items per sub-scale version with a 1–5 rating scale (1=“not at all”, 5=“very much”). We extended the questionnaire such that for each question the participants had to give three an- swers, one for each condition. We also added five ques- tions rating general aspects of the system, again for the different conditions: fun, tiredness, cyber-sickness, in- tuitiveness of control metaphor, and realism of control metaphor.
4. Results for Frontal-View vs. Surround-View
First we compare the questionnaire results and pointing task accuracies of the frontal view (FV) condition to the surround-view (SV) condition.
Since the data from the MEC SPQ was not normally dis- tributed and a log-normal transformation did not solve the
problem, we calculated the results with the non-parametric Mann-Whitney-U test. These computations discovered no significant effects on the sub-scalesAttention,Involvement, andSuspension-of-Disbelief, but very clear ones on the fol- lowing sub-scales:
• Spatial Situation Model(W=26011,p=0.00)
• Spatial Presence Self Location(W=8217,p=0.00)
• Spatial Presence Possible Actions(W=25031,p=0.00) Figure5, left, shows the mean values of all sub-scales. Note that all significant sub-scales directly reflect spatial pres- ence. In contrast, attention, involvement, and suspension-of- disbelief are more related to general presence, not spatial presence in particular. This suggests a distinguished impact of the additional five screens on the participants’ subjectively experienced location.
Analyzing the general system ratings (Figure5, center) with the Mann-Whitney-U test we found the following two effects to be significant:
• Realism(W=923.5,p=0.02)
• Fun(W=807.5,p=0.00)
The differences between the ratings forCyber-sickness,In- tuitiveness of control, andTiredness were not significant.
On the one hand, this is somewhat contrary to the liter- ature claiming a wider field of view causing more cyber- sickness [LaV00,SKHR01]. On the other hand, 135 is al- ready a wide field of view to begin with. Since the control metaphor stayed the same for both conditions it is no sur- prise that we found no significances for the questions related to control.
Since our pointing tasks differed in complexity, we ana- lyzed the following groups of tasks individually:
• from every picture to every other,
• from a specific corridor-picture to every other picture,
• from a specific corridor-picture to another specific one,
• from a specific corridor-picture to the center-picture,
• from the center (1st time) to a specific corridor-picture,
• from the center (2nd time) to a specific corridor-picture.
To avoid distortions due to the different order of sub- experiments per participant, we considered only the data from the first sub-experiment for this analysis (Maze 1, respectively first condition). Since all three groups were similar in gender-ratio, age, VR-skill (FV=2.5, SV=2.1, RU=2.1), and even MRTA-scores (FV=6.3, SV=6.3, RU=6.5), we consider this between-subject comparison of pointing-errors the most robust approach. Since this data was not normally distributed, we again employed the Mann- Whitney-U test. Contrary to our expectations we found no significant differences between any groups of pointing tasks in the FV condition compared to the SV condition. Figure5, right, compares the pointing errors between conditions. Al- though the spatial presence was found to be significantly higher with a surround view, it seems to have no effect on the actual orientation in the same virtual environment.
Involvement Suspension of Disbelief Attention Possible Actions Self Location Situation Model
MEC SPQ Ratings
1 2 3 4 5
Front View Surround View Rotating User
* *
** ** **
Fun Tired Intuitive Sick Realistic
General System Ratings
1 2 3 4 5
** * **
Total Error per Metaphor
Front View
Surround View
Rotating User
0306090120150180
Angular Error
Figure 5: Left: Ratings of MEC SPQ comparing its sub-scales for all three conditions (⇤= significant). Center: Ratings for the general system aspects (⇤= significant). Right: Angular error considering each of the 15 pointing actions per participant in the first maze, showing very similar medians and distributions for all three conditions.
5. Results of Rotating World vs. Rotating User
After showing the effects of an extended field of view, we now describe our results on the effects of user embodiment.
Therefore we compare the results of the surround-view (SV) condition (stable user, rotating virtual space) to the rotating- user (RU) condition (stable virtual world, self-rotating user).
The MEC SPQ ratings (Figure5, left) revealed no sig- nificant differences except for theAttentionsub-scale (W = 17398,p=0.05). Since many participants reported a strong focus on the control mechanism in the RU condition, the questions regarding theAttentionsub-scale may have been misunderstood in the sense that they were answered with re- spect to the attention on the hardware controls and not on the task in the virtual maze. According to our MEC SPQ results, disabling virtual world rotation and instead performing di- rection changes by embodied self-rotations does not have a significant effect on spatial presence.
For the general system questions the participants signif- icantly (W =904.5, p=0.01) distinguished their ratings only on theRealismscale in favor of the RU condition (Fig- ure5, center). This is an important result suggesting the re- moval of the abstraction layer introduced by common con- trol metaphors, such as joysticks or other devices used to rotate the world, instead of having an embodied rotation like we are used to from the real world. However, the re- sults forIntuition of controlare not significant. We argue that this is due to the fact that most participants are young, and even when they stated a low VR-experience they are al- ready somewhat familiar with the rotating world paradigm (e.g., through computer games), but are not familiar with self-rotation. The attention to the unknown control metaphor may have had a negative influence on theIntuitionratings.
Similar to the step from FV to SV, the step from SV to RU yields no significant differences in pointing errors be- tween the conditions for the first maze. In fact, Figure 5, right, shows that the error distributions are nearly the same
Errordependency on VR−skill per Condition
Front View
Surround View
Rotating User
020406080100120
Angular Error
low vr skill high vr skill
Figure 6:The angular error of pointing tasks for corridors just traveled decreases with higher VR-skill for the FV and SV conditions, but is independent from it in the RU condition.
The lines in the box-plots represent median values.
for all three conditions, when looked atin total. This even holds for the second and third sub-experiment.
Like in Section4we examined specific pointing tasks in closer detail, but this time observed a difference in correla- tions between pointing error and VR-skill between the two conditions. Namely, pointing from the end of a corridor back to the central room we found a Spearman correlation factor ofr⇡ 0.35 both for the FV and the SV condition. On the contrary, in the RU condition the correlation factor was nearly zero. Figure6shows the dependency of pointing er- ror on VR-skill for all three conditions. Since self-ratings of VR-skill are error prone and no participant stated himself a VR-skill of 5, we classified participants into low VR-skill (1–2) and high VR-skill (3–4). Though statistically not sig-
nificant, the chart clearly shows that in the conditions where the world was virtually rotated (FV and SV) the error de- creases with increased VR-skill, meaning the user benefited from former VR-experience. In the RU condition, however, VR-skill had no influence on the pointing accuracy. Here participants with low VR-skill had nearly the same error as participants with high VR-skill. We did not find such corre- lations with the MRTA-scores, but people with good mental rotation scores in a paper-and-pencil test are not necessarily good in orientation in an immersive virtual environment.
Since the average VR-skill per group and also the means of the MRTA did not differ much, this observation truly seems to depend on the navigation metaphor only.
Because this effect was not true for arbitrary pointing tasks, user rotation appears not to be beneficial for virtual map learning in global (conceptual) space. However, since it was true for the specific pointing task at the end of a cor- ridor just traveled, we suspect it to contribute to orientation and the ability of locating oneself in the immediate (percep- tual) space.
6. Discussion
Our questionnaire results clearly indicate that expanding the user’s view to a full 360 surround-view enhances the feel- ing of spatial presence. It also improves realism and appar- ently is more fun. In addition to the recorded statements we observed that many participants in the Surround-View con- dition looked behind themselves before a pointing action, al- though they could also have rotated the virtual world in front of them. This furthermore supports the actual use of the dis- plays behind the user. However, the quantitative measures of angular error in the pointing task do not show any improve- ment of accuracy. But this is due to the fact that the involved senses and their stimuli are nearly the same for both the FV and SV conditions. The difference in theperipheralsight of 22.5 (=(180 -135 )/2) for each side may be neglected.
Changing the rotation metaphor from a stable user and a rotating virtual world to a stable virtual world and a self-rotating user further improves significantly on realism.
Though otherwise the questionnaire results for the Rotating- User condition do not significantly vary from the Surround- View condition, they do so compared to the Frontal-View condition for spatial presence(Situation Model, Self Loca- tion, Possible Actions) and Fun. This means that the new control metaphor in RU does not undo the enhancements gained by the transition from FV to SV. For pointing actions along corridors just visited the independence of the point- ing error from the VR-skill in the RU condition suggests an advantage over the FV and the SV condition for immediate self-location, not for map learning in general.
Since our participants were rather young and all some- what accustomed to virtual environments and computer con- trols, we may encounter different results for an elderly group
without any such knowledge. Only a single participant of our study fulfills this criteria. At the age of 56 his median angu- lar errors for the pointing task were 113.3 (FV), 83.4 (SV) and 49.4 (RU). Within-subject differences between condi- tions this big were not observed for any younger participants.
7. Conclusion
In this paper we evaluated the effect of the full horizon- tal surround view and the embodied self-rotation of the OCTAVISsystem on spatial presence and orientation. Our results clearly indicate that both features are indeed benefi- cial in the context the OCTAVISwas designed for: training and rehabilitation of spatial cognition for patients with brain functions disorders. This target group typically consists of elderly people with no prior experience in virtual reality. As a consequence, they benefit from the surround view as well as from the rotating-user control metaphor.
Acknowledgments
The authors are grateful to New Media GmbH for construct- ing the room collection used to build the mazes and also to the fine people of the city of Oerlinghausen for participating in the study. This work was supported by the DFG Center of Excellence EXC 277Cognitive Interaction Technology.
References
[AM90] ALFANOP. L., MICHELG. F.: Restricting the field of view: Perceptual and performance effects.Perceptual and motor skills 70, 1 (1990), 35–45.3
[BIPS12] BRUDER G., INTERRANTE V., PHILLIPS L., STEINICKE F.: Redirecting walking and driving for nat- ural navigation in immersive virtual environments. IEEE Transactions on Visualization and Computer Graphics 18, 4 (2012), 538–545.2
[BWÖ04] BASKAYAA., WILSONC., ÖZCANY. Z.: Wayfinding in an unfamiliar environment: Different spatial settings of two polyclinics.Environment and Behavior 36, 6 (2004), 839–867.4 [CGBL98] CHANCES. S., GAUNETF., BEALLA. C., LOOMIS J. M.: Locomotion mode affects the updating of objects encoun- tered during travel: The contribution of vestibular and proprio- ceptive inputs to path integration. Presence: Teleoperators and Virtual Environments 7, 2 (1998), 168–178.2,5
[DCC97] DARKEN R. P., COCKAYNE W. R., CARMEIN D.:
The omni-directional treadmill: A locomotion device for virtual worlds. InProceedings of ACM Symposium on User interface software and technology(1997), pp. 213–221.2
[DE00] DOGUU., ERKIPF.: Spatial factors affecting wayfinding and orientation: A case study in a shopping mall. Environment and Behavior 32, 6 (2000), 731–755.4
[DP02] DARKEN R. P., PETERSON B.: Spatial orientation, wayfinding, and representation. Handbook of virtual environ- ments(2002), 493–518.2
[DS93] DARKENR. P., SIBERTJ. L.: A toolset for navigation in virtual environments. InProceedings of ACM Symposium on User interface software and technology(1993), pp. 157–165.2
[DZK⇤12] DYCKE., ZELLE., KOHSIKA., GREWEP., WIN- TER Y., PIEFKEM., BOTSCHM.: OCTAVIS: An easy-to-use VR-system for clinical studies. InProceedings of Virtual Reality Interaction and Physical Simulation (VRIPHYS)(2012), pp. 127–
136.2,4
[GM98] GRANTS. C., MAGEEL. E.: Contributions of propri- oception to navigation in virtual environments.Human Factors:
The Journal of the Human Factors and Ergonomics Society 40, 3 (1998), 489–497.2
[HB95] HENDRIX C., BARFIELD W.: Presence in virtual en- vironments as a function of visual and auditory cues. InPro- ceedings of the Virtual Reality Annual International Symposium (1995), IEEE, pp. 74–82.3
[Hol02] HOLLERBACHJ.: Locomotion interfaces. InHandbook of virtual environments technology: Design, Implementation, and Applications(2002), Stanney K., (Ed.), Lawrence Erlbaum Asso- ciates, Inc, pp. 239–254.2
[IRA07] INTERRANTE V., RIES B., ANDERSON L.: Seven league boots: A new metaphor for augmented locomotion through moderately large scale immersive virtual environments.
In Proceedings of IEEE Symposium on 3D User Interfaces (2007).2
[KL04] KNAPPJ. M., LOOMISJ. M.: Limited field of view of head-mounted displays is not the cause of distance underestima- tion in virtual environments.Presence: Teleoperators and Virtual Environments 13, 5 (2004), 572–577.3
[LaV00] LAVIOLAJR. J. J.: A discussion of cybersickness in virtual environments. ACM SIGCHI Bulletin 32, 1 (2000), 47–
56.5
[LDP⇤02] LIN J.-W., DUH H. B.-L., PARKER D. E., ABI- RACHEDH., FURNESST. A.: Effects of field of view on pres- ence, enjoyment, memory, and simulator sickness in a virtual environment. InProceedings of IEEE Virtual Reality(2002), pp. 164–171.3
[Lyn60] LYNCHK.: The image of the city. Harvard-MIT Joint Center for Urban Studies Series. The MIT Press, 1960.2 [MS09] MARQUARDTG., SCHMIEGP.: Dementia-friendly ar-
chitecture: Environments that facilitate wayfinding in nursing homes.American Journal of Alzheimer’s Disease and Other De- mentia 24, 4 (2009), 333–340.4
[MW98] MCCREARYF. A., WILLIGESR. C.: Effects of age and field-of-viewon spatial learning in an immersive virtual en- vironment. Proceedings of the Human Factors and Ergonomics Society Annual Meeting 42, 21 (1998), 1491–1495.3
[NHS04] NITZSCHEN., HANEBECKU. D., SCHMIDTG.: Mo- tion compression for telepresent walking in large target environ- ments. Presence: Teleoperators and Virtual Environments 13, 1 (2004), 44–60.2
[NRI12] NYBAKKEA., RAMAKRISHNANR., INTERRANTEV.:
Are motorized wheelchairs an effective method of locomotion in virtual environments? InIEEE Virtual Reality Short Papers and Posters(2012), pp. 75–76.3
[PFW09] PECKT. C., FUCHSH., WHITTONM. C.: Evaluation of reorientation techniques and distractors for walking in large virtual environments. IEEE Transactions on Visualization and Computer Graphics 15, 3 (2009), 383–394.2
[PI67] PIAGET J., INHELDER B.: The child’s conception of space. W. W. Norton and Company, Inc., 1967.2
[Pso95] PSOTKAJ.: Immersive training systems: Virtual reality and education and training.Instructional science 23, 5-6 (1995), 405–431.2
[Raz05] RAZZAQUES.:Redirected walking. PhD thesis, Univer- sity of North Carolina at Chapel Hill, 2005.2
[RBM⇤10] RIECKE B. E., BODENHEIMER B., MCNAMARA T. P., WILLIAMSB., PENGP., FEUEREISSEND.: Do we need to walk for effective virtual reality navigation? Physical rotations alone may suffice. InProceedings of the 7th international con- ference on Spatial cognition(2010), pp. 234–247.2
[RE98] RAUBALM., EGENHOFERM. J.: Comparing the com- plexity of wayfinding tasks in built environments. Environment and Planning B: Planning and Design 25, 6 (1998), 895–914.2 [RL09] RUDDLER. A., LESSELS S.: The benefits of using a
walking interface to navigate virtual environments.ACM Trans- actions on Computer-Human Interaction 16, 1 (2009), 5:1–5:18.
2
[RSPA⇤06] RIECKE B. E., SCHULTE-PELKUM J., AVRAAMIDESM. N., HEYDE M. V. D., BÜLTHOFF H. H.:
Cognitive factors can influence self-motion perception (vection) in virtual reality. ACM Transactions on Applied Perception 3, 3 (2006), 194–216.2
[RVB11] RUDDLE R. A., VOLKOVA E., BÜLTHOFF H. H.:
Walking improves your cognitive map in environments that are large-scale and large in extent.ACM Transactions on Computer- Human Interaction 18, 2 (2011), 10:1–10:20.2
[SBJ⇤10] STEINICKEF., BRUDERG., JERALD J., FRENZH., LAPPE M.: Estimation of detection thresholds for redirected walking techniques. IEEE Transactions on Visualization and Computer Graphics 16, 1 (2010), 17–27.3
[SGF⇤10] SOUMANJ. L., GIORDANOP. R., FRISSENI., LUCA A. D., ERNSTM. O.: Making virtual walking real: Perceptual evaluation of a new treadmill control algorithm. ACM Transac- tions on Applied Perception 7, 2 (2010), 11:1–11:14.2 [SKHR01] SEAYA. F., KRUMD. M., HODGESL., RIBARSKY
W.: Simulator sickness and presence in a high FOV virtual environment. InProceedings of IEEE Virtual Reality,(2001), pp. 299–300.3,5
[VK78] VANDENBERGS. G., KRUSEA. R.: Mental rotations, a group test of three-dimensional spatial visualization.Perceptual and Motor Skills, 47 (1978), 599–604.5
[VWG⇤04] VORDERERP., WIRTHW., GOUVEIAF., BIOCCA F., SAARIT., JÄNCKEL., BÖCKINGS., SCHRAMMH., GYS- BERSA., HARTMANNT., KLIMMTC., LAARNIJ., RAVAJAN., SACAUA., BAUMGARTNERT., JÄNCKEP.: MEC spatial pres- ence questionnaire (MEC-SPQ). Report to the European Com- munity, Project Presence(2004).5
[WK98] WITMERB. G., KLINEP. B.: Judging perceived and traversed distance in virtual environments.Presence: Teleopera- tors and Virtual Environments 7, 2 (1998), 144–167.3 [WNM⇤06] WILLIAMS B., NARASIMHAM G., MCNAMARA
T. P., CARRT. H., RIESERJ. J., BODENHEIMERB.: Updat- ing orientation in large virtual environments using scaled transla- tional gain. InProceedings of Symposium on Applied perception in graphics and visualization(2006), ACM, pp. 21–28.2 [WNR⇤07] WILLIAMSB., NARASIMHAM G., RUMPB., MC-
NAMARAT. P., CARRT. H., RIESERJ., BODENHEIMERB.:
Exploring large virtual environments with an HMD when physi- cal space is limited. InProceedings of the Symposium on Applied perception in graphics and visualization(2007), ACM, pp. 41–
48.2
[ZDK⇤13] ZELL E., DYCK E., KOHSIK A., GREWE P., FLENTGED., WINTERY., PIEFKEM., BOTSCHM.: OCTAVIS: A virtual reality system for clinical studies and rehabilitation. In Eurographics Medical Prize Papers(2013).2
